AI and workload
Can an AI assistant tell who is overloaded next week?
Yes, if it reads each person’s capacity and planned hours from your planning tool. A list pasted into a chat has no working weeks or time off, so the answer is a guess. In Teamhood, Claude or ChatGPT can read working weeks, time off and planned hours over MCP, but only hours someone planned explicitly.
Last updated: 17 November 2026. By Vidas Vasiliauskas.
What an assistant needs to judge workload
- Capacity per person. Weekly working hours, part-time contracts, vacations and half days.
- Holidays. In Teamhood, working days are Monday to Friday for the whole organization, minus its holidays.
- Planned hours across every project. A person who works on 4 projects is overloaded only if the total is too high.
- The same access as the planner. An assistant works with the role of the person who connected it, so it sees only what that person sees.
Ways to answer the question
| Method | What it reads | What it misses |
|---|---|---|
| A spreadsheet pasted into Claude or ChatGPT | The rows you copied | Changes made after you copied them, and anything not in the sheet |
| Claude or ChatGPT connected to Teamhood over MCP | Working weeks, time off and explicitly planned hours, live | Spread hours |
| The Teamhood schedule view | Each person’s and team’s hours across every project. Overbooked days and weeks turn red. | Work tracked outside Teamhood. You read the view yourself. It does not answer a question. |
From the Teamhood facts page, Teamhood help, and Anthropic and OpenAI help, read 17 November 2026.
Explicitly planned hours and spread hours
In Teamhood, a task’s estimate is spread over its working days. When its dates move, its planned hours move with it. These are spread hours. Hours someone plans for a person by hand are explicitly planned hours.
An assistant over MCP reads only explicitly planned hours. If your team plans with estimates, ask the assistant for a list of tasks and dates, and check the totals in the schedule view.
What a connected assistant can do
MCP (Model Context Protocol) is an open standard for connecting AI assistants to apps. Teamhood’s MCP server connects Claude, ChatGPT, Copilot and Cursor.
- Read and change working weeks, time off and planned hours, within the role of the person who connected it.
- Book planned hours when you ask, for example “Plan 6 hours for Anna on the bridge inspection on Tuesday.”
- Show its changes under your name with an AI label. Changes apply at once.
The assistant sends what it reads, such as staff hours and time off, to the company that runs it. To let an assistant only read, connect it with Teamhood’s read-only address, which offers only the tools that read. Claude allows custom connectors on every plan, one on Free. ChatGPT allows custom MCP servers on Business and Enterprise/Edu.
What to check before you trust the answer
- Working weeks and time off are up to date for every person.
- Holidays are set for the organization.
- Your team plans hours explicitly, or you check spread hours in the schedule view.
- The person who connected the assistant can see every project.
Where Teamhood fits
Teamhood is project management software with Kanban, Gantt and timesheets, for engineering and technical firms that run many projects with the same people. Planned hours count across every project, so you see who has time before you commit them. A task can be given to a team, and its hours count against the team’s capacity, which is the sum of its members’ capacity.
What Teamhood does not do
- An assistant over MCP reads only explicitly planned hours. Spread hours are left out.
- There is no setting to turn off MCP connections for the whole organization.
- No mobile app. Teamhood runs in the browser.
Questions
Can an AI assistant assign tasks based on workload?
Can the assistant change someone’s time off?
Where does the workload data go?
Sources
- Teamhood facts. Capacity planning, MCP server, Hosting and Limits rows.
- Teamhood help: How MCP works in Teamhood and MCP server. Read 17 November 2026.
- Teamhood help: How hour allocation works. Read 17 November 2026.
- Teamhood help: Schedule view. Read 17 November 2026.
- Anthropic help: Get started with custom connectors using remote MCP. Read 17 November 2026.
- OpenAI help: Connecting custom MCP servers in ChatGPT. Read 17 November 2026.